We consider optimal strategies for harvesting a population that is com
posed of two local populations. The local populations are connected by
the dispersal of juveniles, e.g. larvae, and together form a metapopu
lation. We model the metapopulation dynamics using coupled difference
equations. Dynamic programming is used to determine policies for explo
itation that are economically optimal. The metapopulation harvesting t
heory is applied to a hypothetical fishery and optimal strategies are
compared to harvesting strategies that assume the metapopulation is co
mposed either of single unconnected populations or of one well-mixed p
opulation. Local populations that have high per capita larval producti
on should be more conservatively harvested than would be predicted usi
ng conventional theory. Recognizing the metapopulation structure of a
stock and using the appropriate theory can significantly improve econo
mic gains.